Data for Resting Site Preferences in Free-ranging Dogs
Description
The dataset documents resting behavior and site selection of 66 free-ranging dog groups (N = 284) across India over three years (2019–2022) during different seasons. It includes 12 physical parameters of resting sites, behavioral states, and climatic conditions, recorded using a custom Android application. The data reveal that dogs prefer resting sites near resources, at the center of their territories, with maximum visibility, minimal disturbance, and low insect presence, suggesting adaptive strategies for resource defense, thermoregulation, and insect avoidance. This dataset can inform studies on urban animal behavior and aid urban planners in designing inclusive spaces for improved coexistence.
Files
Steps to reproduce
All analyses were performed in R Studio (version 4.2.0). Resting behaviors were categorized into three energy-conservation levels: slightly relaxed, moderately relaxed, and extremely relaxed. To examine the effect of resting site characteristics and climatic factors on the proportion of dogs resting in a group, we utilized generalized linear mixed models (GLMMs) with the lmer package. Resting proportion (resting dogs/total group size) served as the response variable, while site characteristics, temperature, season, and time of day (grouped into four sessions: morning, noon, evening, and night) were predictors, with group identity as a random factor. Box-Cox transformations ensured model assumptions were met, and final models were selected based on Akaike Information Criterion (AIC). Comparative tests (Kruskal-Wallis and Wilcoxon rank-sum tests) were used to analyze physical properties of resting sites across categories. Proportions of individual preferences for specific site characteristics were calculated and compared using Dunn’s post hoc test with Bonferroni corrections. The frequency distribution of unique parameter combinations in selected resting sites was fitted to a Power Law using the poweRlawpackage, indicating non-random selection. These analyses were performed for the entire dataset and separately for each behavioral category.